Rule-Based Embedded HMMs Phoneme Classification to Improve Qur’anic Recitation Recognition

Phoneme classification performance is a critical factor for the successful implementation of speech recognition system. A mispronunciation Arabic short vowels or long can change meaning complete sentence. However, correctly distinguishing phonemes with in Quranic recitation (the Holy book Muslims) still challenging problem even state-of-the-art methods, where duration considered one important features recitation, which called Medd, means that phoneme lengthening governed by strict rules. These call an additional Qur’anic due to based on language characteristics insufficient recognize Tajweed rules, including rules Medd. This paper introduces Rule-Based Duration Algorithm improve recitation. The dataset contain 21 Ayats collected from 30 reciters and are carefully analyzed baseline HMM-based model. Using Hidden Markov Model tied-state triphones, set models optimized constructed integrated into method. proposed algorithm achieved outstanding accuracy, ranging 99.87% 100% according Medd type. obtained results will contribute significantly models.